Star pattern recognition based on features invariant under rotation

Abstract

A star pattern recognition algorithm is proposed on the basis of features invariant under rotation. Guidance star identification via the algorithm is performed on star images captured by star sensor and simulated images. The results indicate that the proposed method presents better robustness against position and magnitude noise than conventional ones and eliminates rotation procedure and avoids the influence caused by grid size choice. The database feature storage for each pattern consists of simply four floating point numbers.